Skip to Content

Homomorphic Encryption

Making cloud computing more secure
April 19, 2011
Ciphering: Gentry’s system allows encrypted data to be analyzed in the cloud. In this example, we wish to add 1 and 2. The data is encrypted so that 1 becomes 33 and 2 becomes 54. The encrypted data is sent to the cloud and processed: the result (87) can be downloaded from the cloud and decrypted to provide the final answer (3).Ciphering: Gentry’s system allows encrypted data to be analyzed in the cloud. In this example, we wish to add 1 and 2. The data is encrypted so that 1 becomes 33 and 2 becomes 54. The encrypted data is sent to the cloud and processed: the result (87) can be downloaded from the cloud and decrypted to provide the final answer (3).
Ciphering: Gentry’s system allows encrypted data to be analyzed in the cloud. In this example, we wish to add 1 and 2. The data is encrypted so that 1 becomes 33 and 2 becomes 54. The encrypted data is sent to the cloud and processed: the result (87) can be downloaded from the cloud and decrypted to provide the final answer (3).

Craig Gentry is creating an encryption system that could solve the problem keeping many organizations from using cloud computing to analyze and mine data: it’s too much of a security risk to give a public cloud provider such as Amazon or Google access to unencrypted data.

The problem is that while data can be sent to and from a cloud provider’s data center in encrypted form, the servers that power a cloud can’t do any work on it that way. Now Gentry, an IBM researcher, has shown that it is possible to analyze data without decrypting it. The key is to encrypt the data in such a way that performing a mathematical operation on the encrypted information and then decrypting the result produces the same answer as performing an analogous operation on the unencrypted data. The correspondence between the operations on unencrypted data and the operations to be performed on encrypted data is known as a homomorphism. “In principle,” says Gentry, “something like this could be used to secure operations over the Internet.”

With homomorphic encryption, a company could encrypt its entire database of e-mails and upload it to a cloud. Then it could use the cloud-stored data as desired—for example, to search the database to understand how its workers collaborate. The results would be downloaded and decrypted without ever exposing the details of a single e-mail.

Gentry began tackling homomorphic encryption in 2008. At first he was able to perform only a few basic operations on encrypted data before his system started producing garbage. Unfortunately, a task like finding a piece of text in an e-mail requires chaining together thousands of basic operations. His solution was to use a second layer of encryption, essentially to protect intermediate results when the system broke down and needed to be reset.

“The problem of how to create true homomorphic encryption has been debated for more than 30 years, and Craig was the first person who got it right and figured out how to make the math work,” says Paul Kocher, the president of the security firm Cryptography Research. However, Kocher warns, because Gentry’s scheme currently requires a huge amount of computation, there’s a long way to go before it will be widely usable.

Gentry acknowledges that the way he applied the double layer of encryption was “a bit of a hack” and that the system runs too slowly for practical use, but he is working on optimizing it for specific applications such as searching databases for records. He estimates that these applications could be ready for the market in five to 10 years.

Keep Reading

Most Popular

mouse engineered to grow human hair
mouse engineered to grow human hair

Going bald? Lab-grown hair cells could be on the way

These biotech companies are reprogramming cells to treat baldness, but it’s still early days.

Death and Jeff Bezos
Death and Jeff Bezos

Meet Altos Labs, Silicon Valley’s latest wild bet on living forever

Funders of a deep-pocketed new "rejuvenation" startup are said to include Jeff Bezos and Yuri Milner.

ai learning to multitask concept
ai learning to multitask concept

Meta’s new learning algorithm can teach AI to multi-task

The single technique for teaching neural networks multiple skills is a step towards general-purpose AI.

conceptual illustration showing various women's faces being scanned
conceptual illustration showing various women's faces being scanned

A horrifying new AI app swaps women into porn videos with a click

Deepfake researchers have long feared the day this would arrive.

Stay connected

Illustration by Rose WongIllustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.